基于独立分量分析和遗传算法的图象分离方法研究与实现

在深入分析独立分量分析技术的基础上，针对常规数值求解方法容易陷入局部最优解的问题，提出了一种基于遗传算法和独立分量分析相结合的盲源分离新算法．通过对图象信号分离仿真试验表明，采用最佳保留机制和移民方式的动态补充子代个体操作，在一定的群体规模和遗传代数的情况下，该方法能实现信号的盲分离，并可获得全局最优解．对超高斯信号和亚高斯信号的混合信号，与扩展信息最大化方法相比，该方法可获得更好的分离效果。

Research & Realization of Image Separation Method Based on Independent Component Analysis & Genetic Algorithm

A novel Blind Source Separation(BSS) algorithm based on the combination of genetic algorithm andIndependent Component Analysis (ICA) is proposed with analysis to the ICA method. The proposed algorithm can be used to solve the problem of local optimum that is easily stacked into by normal numerical solution. In the genetic algorithm, the Kurtosis as the fitness function is adopted, the elitist model is introduced and supplying filial generation's individual with migrant operation dynamically is also adopted. The simulation 1 is the separation of the mixed signals of three images and a noise. The simulation 2 is the separation of the mixed signals of two image signals (sub-gauss signal) and two voice signals (super-gauss signal). The image separation simulation shows that the blind signals separation can be realized and the global optimum can be acquired through the proposed algorithm under the circumstance of adequate population size and genetic generations. Compared with the Blind Source Separation method of extended-infomax, the proposed method in this paper can acquire better separating effect in separating the mixed signals of sub-gauss signal and super-gauss signal.